Mainnet Merge Clients Launch – Trustnodes

All major clients of ethereum have launched the Merge ready client for the mainnet upgrade to Proof of Stake (PoS).

The biggest, Geth, needs a small re-do because of some new online pruner, but Besu on eth1 as well as Lighthouse and Prysm on eth2 are all good just two weeks ahead of the beginning of the hardfork.

“All mainnet users must upgrade to this release,” Lighthouse said. “Users who fail to upgrade their nodes before the Bellatrix fork (September 6th) will stop following the canonical chain.”

So the show starts on September 6th, and then enters some sort of purgatory limbo for over a week until the equivalent of the block number kicks in expected on or around September 15th.

That makes this the first upgrade ever to use not a block number for the fork upgrade but a Terminal Total Difficulty of 58_750_000_000_000_000_000_000, or 58,750 with no dollar sign, except on actual terminals.

The Beacon chain will upgrade first on PoS, and that’s at a slot height, on September 6th. Then once total difficulty hits this 58,750 or above, the Proof of Work network will transform into Proof of Stake with network difficulty at that point set to zero because there will no longer be any Proof of Work mining with its difficulty orchestration.

Where node operators are concerned, therefore, they have to be on these new upgrade clients within two weeks, rather than by September 15th.

That applies to all node operators of any variety because a lot of things can potentially happen after September the 6th to fork you off the network if you’re still running non-upgraded software.

The most talked about is a delay of total difficulty, but in theory one could speed it up by adding a lot of hash to fork on say September 7th rather than on the expected 15th.

A block number would guard from such volatility as you have to go through all the block numbers first and difficulty is meant to stabilize those new block numbers to about 12 seconds for eth.

The only thing that determines difficulty however is just how much hash you put in, and so this ‘attack’ vector to fork maybe even on September the 6th itself is not only potentially viable, but it has actually been tested on the Ropsten testnet by unknown actors.

That does not imply there are plans to utilize it, but it does suggest that at least ‘vital’ nodes must upgrade by September 6th.

The opposite might instead happen whereby miners vainly try to delay total difficulty hitting this magic 58,750 number by simply withdrawing hash.

Miners however will keep receiving actual and valuable eth until the very last block below 60k-ish difficulty, and so pure self interest would break any such attempt in addition to it being quite easy to get gamers to GPU the merge, not to mention the Ethereum Foundation has said they will hire hash if necessary.

Which is probably why the delay vector has not been tried, but the speed up vector has been tested, not least because just delaying it a bit wouldn’t cause any mess – to the contrary the ecosystem would be even more prepared with even super laggard nodes upgrading – but a fork on the 7th when potentially some ordinary nodes think they have time until say the 12th of September at least, can cause a mess and did cause a mess on Ropsten.

Eth devs had to do over that test-net upgrade because they were taken by surprise at its activation way sooner than they thought.

The main-net however won’t be taken by surprise, or at least shouldn’t, because the nodes that matter should now know what can potentially happen.

Those nodes should act as if the testnet attack was or is actually a plan for a mainnet attack, and so all must upgrade by September 5th, while where the entire ecosystem is concerned realistically we think it was probably a dev, and maybe an eth dev rather than a bitcoin dev, trying to make a point that this difficulty thing isn’t how you do upgrades when you have block numbers.

And so the total difficulty on your terminal might be the first, and probably last time, we see an upgrade using this mechanism, but the merge is bringing something that might be a lot more long term useful called PREVRANDAO.

We get DAO, and rand is random. Prev? Hopefully no joke about previously random and no more, because this new parameter prints “the output of the randomness beacon provided by the beacon chain.”

“Applications may benefit from using the randomness accumulated by the beacon chain. Thus, randomness outputs produced by the beacon chain should be accessible in the EVM,” the EIP says.

Provably fair poker therefore is getting a $200 billion bounty to prove that it is provably fair because if you can game the Beacon randomness, then you get to be the proposing staker all the time, or the equivalent of the miner that finds all the blocks on his or her windows 95 CPU mining.

This new capability has far more use than just poker, but also makes the Beacon randomness probably the most valuable thing in all of eth, and maybe even all of crypto, and thus may well be where any weakness may lie.

On the other hand exploiting any such weakness just means it will be fixed, and fixing randomness gets you to frontier maths, which thus translates to a wider scientific advancement.

In practice that may well mean higher efficiency at getting a cool unique avatar, or on-chain art, but randomness goes towards understanding entropy and many other things, so this Prevda is the coolest thing on the Merge after the miner slashing, the issuance certainty, its potentially deflationary nature, and of course the world’s first major yield bearing crypto asset.

And all this now starts in just two weeks with the full upgrade being pretty much on the day summer ends, signifying a new beginning.